HyperTransport and Opteron/Athlon64 Overclocking
The first question many will have about our efforts to look at how Athlon64 will perform is how we can possibly compare an overclocked Opteron to a chip that is not overclocked. In the case of the Opteron, the comparison is more accurate than you might first think.In normal setups (e.g. Athlon/P4), the CPU gets its clock from the FSB clock and multiplies it by the “clock multiplier” to determine how fast its internal clock should be. With a 16x multiplier, when the external clock ticks once, the CPU ticks 16 times. However, with the Athlon 64/Opteron, there is no FSB, so the CPU must get its clock from somewhere. It doesn't produce it internally; instead, it derives it from the native HT (HyperTransport) frequency, which is 200MHz, but because of the bus' nature, it runs at an effective 800MHz.
So, for our 1.8GHz Opteron 144, the multiplier is 9x, which is why raising the HT frequency to 222MHz increases the clock speed to around 2GHz. But we are increasing the HyperTransport clock in our overclocking, and not a FSB clock, which does not exist on Opteron/Athlon64. In real terms, this means our CPU overclocking has a significant impact on Performance, but it is unlikely that our increase in memory speed will have nearly as much impact on performance. Since we are nowhere near saturating the Hypertransport bus at 200 (effective 800), increasing HyperTransport to 222 (888) will not likely have much, if any, impact on overall performance. Our performance improvements, with Opteron/Athlon64, are mainly coming from increase in CPU clock — much more so than on the Pentium 4 or Athlon architectures.
Obviously, the PCI bus operates at a different frequency than the HT bus than the CPU, but they all operate based on multiples of each other, and are all derived from the HyperTransport clock.
Facebook
Twitter
RSS
79 Comments
View All Comments
Anonymous User - Friday, September 5, 2003 - link
If you're going to compare an overclocked Opteron to the Pentium 4, shouldn't you at least use the fastest P4 available (3.2ghz vs 3.0) ?Anonymous User - Friday, September 5, 2003 - link
LOL, you mean Onlt = only? whilee = while and 22Hz = 222Mhz=Anonymous User - Friday, September 5, 2003 - link
You fools! Onlt the HT-bus was overclocked not the memory bus! You guys need to educate about hammer systems. I sugest you get yourself and nf3 then you can see that you can overclock the HT bus and not the memory!! The test is very valid cause the 2Ghz is the real frequency whilee the 22Ghz is plain bus HTB speed and has nothing to do with the memory controller or memory bus speed.dvinnen - Friday, September 5, 2003 - link
#39: You twit, I never said that the HT line had anything to do with the memory. It is used for the clock of the computer. THat was what they where ocing, not the memory.#44: I also dought there will be that mch difference in proformance between the dual channel and single channel. Just as the current athlon dosen't really beniffent from the extra bandwidth, I dought the Athlon64 will. And the fact that this is an ethuist website is exactly why they should be doing a preview of it.
Anonymous User - Friday, September 5, 2003 - link
Can I please have benchmarks under linux compiled in 64-bit mode :). I disagree with your "we will have to wait for 64-bit windows" philosophy. :)Anonymous User - Friday, September 5, 2003 - link
There is either a lot of AMD bias, ignorance, or both in this article. There is no way that the performance of an upcoming true Athlon64 can be compared to a current overclocked Opteron. The Athlon64 is going to be a socket 754 single channel DDR400 solution that will barely keep up with an Athlon XP 3200+ let alone a P4 3.2 GHz or upcoming Prescott 3.4 GHz. The author should have compared the overclocked Opteron in his article to an upcoming Athlon FX which is just a remarked socket 940 Opteron that runs dual channel DDR400 ECC registered memory. The only difference is in price - which seems funny that is is never mentioned in articles where AMD wins but when Intel wins it's all about COST!!! Athlon64 socket 754 is expected to be price competitive with current Athlon XP 3200+ since the Athlon64 will initially be marked either 3100+ or 3200+. However, Athlon FX is expected to sell anywhere from $700 - $900 USD or even more at launch. So once you buy the 6-layer motherboard, the ECC registered DDR400 memory, and the Athlon FX, you'll have paid almost twice as much for a system that is almost "faster than a P4 3.2 GHz in every way" as some Fudo person stated it on the inquirer.net. I'm collecting a compilation of all of these hype articles to send back to the authors after the real launch when the real truth is known. AMD has been promising that the 64-bit Hammer was going to be the next best thing to sliced bread for the last 3 years and that the Athlon64 was going to wipe Intel off the face of the earth for the past year or so and they still haven't delivered. Everyone who is waiting to get an Athlon FX better pre-order one pretty quick because there will probably only be about 10,000 available world-wide this year.Anonymous User - Friday, September 5, 2003 - link
Reply to #41It is nice to support Athlon 64, but dont tell people that it would recieve an 1 MB L2 cache, as AMD is already building 512 KB CPU's to ship on the launch date.
And if gunmetal was a good DX9 benchmark, my FX 5200 would be a good R 9800 alternative, and you can see in other reviews that this is not so.
Anonymous User - Friday, September 5, 2003 - link
And #39, you're stupid. They overclocked the FSB to get it to 2.0GHz while keeping the memory at DDR400. You can do that with modern boards and you can certainly do it with A64 boards or Opteron boards.Get a clue.
Anonymous User - Friday, September 5, 2003 - link
#40, get a clue. The enthusiast edition of the Athlon 64 will have the exact same L2 cache as the Opteron (1MB). In fact, this same high-end Athlon 64 will have a higher clock speed (probably 2.2GHz) and will be identical to the Opteron. Therefore, this review is VERY useful.And no, you're wrong about Gunmetal, it's a very good DX9 benchmark.
Based on these facts, this review gave the A64 exactly the credit it's due.
Anonymous User - Friday, September 5, 2003 - link
A point that I think was overlooked in this review is that the Athlon 64 has a lot less L2 cache than the tested Opteron, about a P4 to Celeron difference. So you might be very dissapointed when you actually buy an Athlon 64, when you compare it to the review.Besides, the best score was made in an added benchmark, gun metal, that seems to use some kind of software rendering for 3d, a very unrealistic scenario. I think so because my Geforce FX 5200 scores almost the same as the tested Radeon 9800, as I have one of the tested CPU's.
And another point of intrest is the video editing, that is hardly covered and does not even get its own part in the review, altrough this might be the only point where consumers actually use their full CPU power, as gaming is more a GPU thing, and anything above 60 fps does not make it more fun than just 60 fps. And it is also the only point where the Opteron seems to lack to most other CPU's.
Based on my other points, I have to conclude that Athlon 64 got more credit here than it deserved.